Improve error messages / documentation around scanning hive directories

[nameexhaustion]

Description

Some hive datasets don't work out of the box if the directory is passed to scan_parquet because they contain non-data files, and currently the error messages we end up printing are very cryptic (e.g. parquet: File out of specification: The file must end with PAR1). We should check the file extensions and raise a better error message if we see mixed extensions.